Measurements of U.S. rivers clarify river-shaping factors and interaction with groundwater



Journal Title

Journal ISSN

Volume Title



We strive to answer the following: Why do modern rivers have the morphology (width and sinuosity) that they do, and how does morphology influence water quality? Study 1: We used the National Hydrography Dataset (NHD) to measure river widths nationally. We find that our measured median widths are controlled primarily by drainage area (r = 0.73). Across the contiguous U.S., width standard deviation within a given river reach averages 25% of mean width (r = 0.92), i.e. variability is ~2x. Study 2: We used the National Hydrography Dataset (NHD) to measure river sinuosities nationally. Sinuosities were measured over reaches scaled by river width. Measured sinuosities typically increase with increasing reach length. By using short reaches of ~1 meander arc length, we aim to isolate sinuosity of channels meandering within floodplains. We found that remote sinuosity measurements across the eastern half of the contiguous U.S. exhibit the following correlations: drainage density r = -0.34; land slope r = -0.37; soil sand content r = 0.54. Sand-rich soil is a key driver of high sinuosity due to a self-reinforcing mass balance cycle. Outer banks consisting of sand-rich soil are eroded and thus contribute sand to the river bed, sustaining point bar lateral accretion, forcing high-velocity flow core toward the outer bank, and thus further eroding the bank. Our study provides the first evidence of this feedback loop at a broad spatial scale. Study 3: We use sinuosity and other data to remotely calculate hyporheic exchange--here, river-groundwater through-bank exchange--along all mapped NHD rivers within the Ohio, Upper Mississippi, and Missouri River Basins. We calculate that 99.6% of all water reaching the confluence of the Ohio and Upper Mississippi Rivers has entered the hyporheic zone at some point. The mean and median flow length necessary for 50% exchange to occur are 1.64 and 0.67 km, respectively. We find that only 24.2% of river length would be expected to exhibit net denitrification.